INDEX
Explanations
racial demographics
This neuron fires on decimal‐formatted numeric tokens—especially the fractional percentage values in demographic/statistics sections.
New Auto-Interp
Negative Logits
áreas
-0.06
(sys
-0.06
َال
-0.06
�
-0.06
_ts
-0.06
_STORAGE
-0.06
همسر
-0.06
legal
-0.06
ої
-0.06
Jail
-0.06
POSITIVE LOGITS
거야
0.07
�
0.07
галтер
0.06
Coconut
0.06
Ultra
0.06
Archie
0.06
исключ
0.06
unker
0.06
числі
0.06
initWith
0.06
Activations Density 0.001%